I don't know, this somehow seems to ignore the case when chapters rely on something that should be clear from previous chapters. It happened to me a lot when I was at University that I'd think to just skim over a chapter, poorly understand the following one and that would snowball as soon as I reached the end of the book.
That would mean you went deeper in this chapter than in the previous one. The first fast iteration is really to get familiar with the main ideas and terminology. For example if it's a book on algorithms, you skim through it and understand what is linkedlist, what is tree, what is graph, their running time but not the details of implementation and certainly not exercises. Then on the next one you try to understand basic implementation, but still not doing any exercises. Then on the next one you try to implement them and do some exercises. That's fun, keep you moving, refreshing your memory and it doesn't feel like you've been reading this book forever and still on the 3 chapter.
Well, that's when you try to use skimming as a foundation for actually learning the next concept. If I'm just trying to get an outline, all I really have to understand is that -for example- the Jungian and Freudian models of the mind are very different, but were foundational for much of psychology.
On the next pass, I learn more about what they are, and I retain it better because I remember the skimmed materials from later lessons that relate to particulars of those psychological models.